This study was designed to assess long-term retention following either keyword learning or rote rehearsal under conditions of intentional learning. A 2 condition (keyword vs. rote rehearsal) x time (immediate vs. delay) completely randomized design was used, with 24 Tagalog language nouns serving as the learning material. All subjects had uniformly high levels of retention on the immediate test of cued recall. However, on the delayed test of cued recall (2 days later), keyword learners had forgotten almost twice the number of items as had subjects who had rehearsed the same material by rote. Long -term forgetting was also greater for the keyword condition than for rote rehearsal when subjects were assessed by a test of associative matching. This pattern of results replicates earlier research that has revealed a detrimental long-term effect ofthe keyword mnemonic under conditions of incidental learning.Considerable evidence exists for the beneficial effect of the keyword mnemonic on acquisition and immediate recall. Although originally developed by Richard Atkinson (1975) as a study strategy for second language vocabulary learning, its usefulness has been documented across a wide range oflearning tasks. The acquisition of science concepts, art history, the names of state capitals, and factual information presented in a prose format are but a few examples of its broad applicability (Carney, Levin, & Morrison, 1988;Levin, Shriberg, Miller, McCormick, & Levin, 1980;McCormick, & Levin, 1987;Pressley, Levin , & Delaney, 1982; Rosenbeck, Levin, & Levin, 1989).Whereas the immediate advantage of using keywords is well documented, the possible long-term benefit of this mnemonic technique has not been established. In fact , some evidence suggests that the keyword mnemonic may be associated with a relatively steep decline in retention if recall is assessed after a delay (McDaniel, Pressley, & Dunay, 1987;Wang, Thomas, & Ouellette, 1992) . conducted four experiments to compare long-term retention after keyword learning with that after rote rehearsal . We were somewhat surprised to find that after 1 week, greater forgetting occurred under use of the keyword mnemonic than under rote rehearsal.In the aforementioned experiments, condition (keyword vs. rote rehearsal) and time (immediate vs. delay) were factorially manipulated as between-subjects variables . Despite variations in the study materials (French or Tagalog language nouns), and levels ofinunediate recall (equated or not equated across learning conditions), keyword learners typically forgot more than twice the number of vocabulary items than did subjects who rote rehearsed the same material. In each of these experiments, a highly significant condition X time interaction was obtained because keyword learners were less likely to retain as much after a delay as rote learners. Moreover, the poor long-term performance of the keyword learners was not due to their forgetting of the keywords . For instance, after a l-week delay , keyword recall was 90% , as opposed to a meage...